A problem faced by many Web services and other applications today is to achieve very high scale and very high availability at the same time, while keeping operational and capital costs low. This problem is particularly challenging when the application needs to dynamically update stateful databases such as SQL databases. For stateful databases, at high volumes of data and/or high transaction rates, having many identical copies of all the data is not technically or economically feasible. The brute force approach of “all data everywhere” would consume too much disk space, and transaction rates would be too rapid for disk, server, and network bandwidth.
Hence, each piece of data must be located on only a small subset of servers, and an application must route each request to one of that subset of servers that has the desired data, plus handle failover to another server if the first choice is not available. In addition, there are significant operational challenges to monitor and administer large numbers of middleware and backend servers.
In view of the foregoing, there is a need for systems and methods that overcome the limitations and drawbacks of the prior art.